R E S E A R C H A R T I C L E
Erosion and sediment transfer processes at the front of rapidly
moving rock glaciers: Systematic observations with automatic
cameras in the western Swiss Alps
Mario Kummert
|
Reynald Delaloye
|
Luc Braillard
Department of Geosciences/Geography,University of Fribourg, Switzerland Correspondence
Mario Kummert, Department of Geosciences/ Geography, University of Fribourg,
Switzerland.
E‐mail: [email protected]
Abstract
When connected to torrential channels, the fronts of active rock glaciers constitute important sediment sources for gravitational transfer processes. In this study, a 2013–16 time series of in situ webcam images from the western Swiss Alps was analyzed to characterize the erosion pro-cesses responsible for sediment transfer at the front of three rapidly moving rock glaciers and their temporal behavior. The main erosion processes comprised rock fall, debris slide, superficial flow and concentrated flow. These processes were induced by (i) changes of the frontal slope angle produced by rock glacier advance, and (ii) increases in water content of the sediments at the rock glacier front due to melt processes and rainfall. Erosion almost ceased during winter, when the front was frozen and snow‐covered. The onset of snowmelt triggered an active period of high‐frequency erosion events. After the melt period, sediment transfer continued as occa-sional rock falls, while other erosion processes occurred only during or following rainfall events. Intense regressive erosion phases that triggered debris flows were rare and occurred when enhanced snowmelt and/or recurring rainfall induced substantial groundwater flow on the debris slopes directly below the rock glacier fronts.
K E Y W O R D S
debris flow, erosion processes, rock glacier, sediment transfer, temporal variability
1
|I N T R O D U C T I O N
Rock glaciers on mountain slopes can constitute a significant source of loose debris available for gravitational transfer processes within the sediment cascade.1The efficiency of the transport defines the level
of connectivity between the different compartments of the cascade (eg,2) and strongly relies on both the presence of uninterrupted steep
slopes and channels guiding the sediments downslope, and the avail-ability of unconsolidated sediments. Debris delivered to the rock gla-cier front can be mobilized further downslope, mainly via rock falls, surface runoff and debris flows. As many rock glaciers have acceler-ated considerably in recent decades—for example in the European Alps (eg,3–6), the Brooks Range in northern Alaska7and the Kazakh and
Kyrgyz Tien Shan8,9
—the availability of unconsolidated sediments
downslope may have increased substantially, potentially affecting the sediment transfer rate within the sediment cascade.
The sediment supply from the upper part of mountainous water-sheds strongly influences the development of torrential hazards but its study is commonly neglected.10There is a general lack of
informa-tion about erosion processes in the upper secinforma-tions of torrential catch-ments, as hazard assessment is often limited to observations on the alluvial fans where most infrastructure is located (eg,11,12). Identifying sediment storage and active debris inputs in the headwaters of high alpine catchments is essential to assess the magnitude and frequency of debris flow events (eg,13–15). Along with the quantification of sed-iment fluxes16, understanding what controls erosion from the main sediment sources is the key to estimate sediment budgets17 and to
identify debris flow scenarios.18
-This is an open access article under the terms of the Creative Commons Attribution‐NonCommercial‐NoDerivs License, which permits use and distribution in any
medium, provided the original work is properly cited, the use is non‐commercial and no modifications or adaptations are made.
© 2017 The Authors. Permafrost and Periglacial Processes published by John Wiley & Sons, Ltd. DOI: 10.1002/ppp.1960
Together with other sediment sources (eg, moraines, talus slopes and weathering accumulations), rock glaciers can occur in catchments located in mountainous periglacial terrain. In the case of a rock glacier advancing over gently inclined terrain, the sediments mobilized from the front by gravity accumulate at its foot, creating a small talus slope that is soon overridden by the rock glacier.19–21 As a result, a stiff blocky layer forms at the base of the deforming ice‐rich permafrost (eg,22) and the sediments cannot be mobilized further downslope by debris flows or other gravitational processes (type A on Figure 1). Thus, rock glaciers of type A represent traps in the sediment cascade23–25, as there is no sediment connectivity with further down the slope. In con-trast, an efficient connectivity occurs when the front of an active rock glacier reaches a torrential channel or a steep, often convex slope (type B on Figure 1). The sediments eroded from such a rock glacier front create what we define here as a debris slope, ie, a relatively large depo-sitional area consisting of a mixture of fine to coarse rock debris (Figure 1), where sediments can be remobilized later, for instance by debris flows. For the largest boulders, a rock fall with travel distance of tens to hundreds of meters may even be triggered directly from the rock glacier front (Figure 2). Although type B rock glaciers appear to be less frequent than type A ones21, several examples of the former
have been observed in the European Alps (eg,26,27). However, few studies have focused on the erosion of the front of active rock glaciers (28,29for type A; and30for type B) or on the relation between rock gla-ciers and debris flow occurrence (eg,26,31). To our knowledge, no study
has detailed the processes eroding rock glacier fronts, their temporal behavior and their controlling factors.
The present study aims to (i) identify and classify the processes that erode rock glacier fronts that are efficiently connected to torrential gullies or steep slopes, (ii) investigate the temporal variability of their dynamics and (iii) identify their controlling factors. For this, an extensive database of in situ webcam images has been collected for three study sites located in the Swiss Alps. Images from January 2013 to December 2016 were qualitatively analyzed and interpreted to reveal both the types and the temporal behaviors of erosion processes occurring at the front of the rock glaciers and within the uppermost part of their connected torrential systems (“area of interest” on Figure 1).
2
|S T U D Y S I T E S
The three investigated rock glaciers (Dirru, Gugla and Tsarmine) are located in the southwestern Swiss Alps (Figure 3). They all face west, with a terminus developing over a steep convex slope and dominating a torren-tial gully, allowing an efficient sediment connectivity with torrents. The adjacent torrential channels collect water from contributing areas of 0.11 km2at Dirru, 0.64 km2at Gugla and 0.27 km2at Tsarmine. None
of the three sites displays an organized surface runoff network upslope from the rock glacier front. Along with their topographical setting, the rock glaciers were selected because of their high current flow rate, favor-ing a substantial sediment transfer between the fronts and the torrential gullies, and thus allowing observations to be made within only a few years. During the documented period (January 2013 to December 2016), each rock glacier was characterized by mean displacement rate of several meters per year (m/y) (Figure 4). In addition, webcam images showed that the position of the respective front lines, ie, the erosion bor-der of the rock glacier surface (Figure 1), did not change much during the study period, while the total movements of the rock glaciers are estimated to have reached approximately 28 m at Dirru, 18 to 53 m at Gugla (18 m for the northern front and 53 m for the southern front) and 17 m at Tsarmine. Therefore, the advance of the three rock glaciers must have been balanced approximately by erosion of their fronts.
2.1
|Dirru
The Dirru rock glacier (46.12° N, 7.81° E) is located in the Mattertal valley (Figure 3). The elevation of the active tongue ranges from 3000 to 2530 m a.s.l. The lower half of the tongue is more than 350 m long and about 50 m wide. The front is located approximately 150 m upslope on the southern side of the Geisstriftbach torrential channel. The disrupted snow cover observed at a specific horizon of the front in winter indicates the presence of one main shear horizon, ie, the horizon where most of the deformation occurs32, located ~15 m below the front line. During the past 10 years, the terminal part of the rock glacier tongue has moved continuously at an annual veloc-ity of ~6–8 m/y.27The bedrock source of the rock glacier is mainly
composed of mica‐rich gneisses and schists.33The latter consists of a heterogeneous mix of coarse debris, including boulders of up to sev-eral cubic meters and a substantial amount of fine‐grained sediments (heavily weathered debris) that are particularly visible at the front. Around 1995, a large erosion niche (~230 m long, 35 m wide and up to 20 m deep) developed downslope from the front, strongly
FIGURE 1 Two scenarios that schematically show the connectivity
between an active rock glacier and the downward slope. A—No connectivity: The sediments are stored at the foot of the front and will be overridden by the advancing rock glacier; B—Efficient connectivity: The sediments are leaving the rock glacier system. The area of interest corresponds to the front and—at least—the upper part of the adjacent debris slope [Colour figure can be viewed at wileyonlinelibrary.com]
FIGURE 3 Images of the three study sites indicating the rock glacier outlines (white dashed lines), their front lines (white dotted lines), the location of in situ webcams (black and white dots) and examples of webcam images [Colour figure can be viewed at wileyonlinelibrary.com]
FIGURE 2 Images of long travel distance rock fall events triggered from the front of the Tsarmine rock glacier in July 2015 (left) and from the front
of the Grabengufer rock glacier (Mattertal, Switzerland) in June 2009 (right). Both falling boulders exceeded 10 m3in volume and traveled several
enhancing the connectivity between the rock glacier and the torrential channel. The niche has been progressively infilled by debris supplied by the rock glacier since then. Small debris flows reaching the Geisstriftbach main channel were triggered in the niche during the snow melt season in June 2013, 2014 and 2016.
2.2
|Gugla
The Gugla (also named Gugla‐Bielzug or Breithorn) rock glacier (46.13° N, 7.81° E) is located about 2 km north of the Dirru rock glacier, on the same valley flank (Figure 3). The geological setting is very similar to Dirru but the morphology, the topographical setting and the dynamics differ. Gugla has a tongue‐shaped morphology and flows toward the west in a small valley between the Breithorn summit on the north side (3178 m a.s.l.) and the Gugla summit on the south (3377 m a.s.l.). The tongue is about 450 m long and 100 m wide, and ranges in elevation between 2820 and 2600 m a.s.l. The terminal section of the rock gla-cier can be divided into a southern part, which is steeper, thicker (20–30 m) and contains several superimposed shear horizons (the disrupted snow cover observed on webcam images from winter 2011/12 and 2012/13 indicated three superimposed shear horizons); and a northern part, less steep with apparently only one shear horizon located about 15 m below the front line. Since 2008 at least, the sur-face velocities have been continuously faster on the southern part (~ 10 m/y) than on the northern part (~ 4 m/y; Figure 4).27Both parts flow
directly into the Bielzug torrential gully, supplying it with rock debris. Such efficient sediment connectivity has been continuously active since at least 1930 (as observed on aerial photographs). Each year from 2012 to 2016, one to several debris flows triggered from an area imme-diately downslope from the rock glacier front reached the main valley close to the village of Herbriggen (St Niklaus, VS). The volumes involved ranged from 500 to more than 5000 m3per event.18
2.3
|Tsarmine
The Tsarmine rock glacier is located about 25 km west of the two other sites (46.04° N, 7.50° E), in the Val d'Arolla (Figure 3). It is tongue‐ shaped, about 450 m long, 100 m wide and ranges in elevation between 2680 and 2460 m a.s.l. The rock glacier surface has an openwork
structure of gneiss boulders up to several cubic meters in size. A matrix of finer sediments—apparently coarser than at the two other sites—is visible at the front, and probably underlies the openwork structure. The main shear horizon is approximately 15 m below the front line according to webcam images in winter (Figure 5). The rock glacier ends on the top of a deep torrential gully where debris has accumulated since at least 1946 (oldest available aerial photograph). During the last 5 years, the annual rock glacier velocity has continuously accelerated from about 1 m/y in 2011 to more than 5 m/y in 2016 (Figure 4).6No evi-dence of debris flow triggered from the front or adjacent debris slope and reaching the main valley has been observed since 1946.
3
|M E T H O D S
The erosion processes linking the front of the rock glaciers to the adja-cent debris slopes were characterized from year‐round time series of webcam images obtained from January 2013 to December 2016, complemented by some direct in situ observations. The approach relies on (i) acquiring a time series of images and (ii) visually identifying ero-sion processes from them.
The three sites were equipped with one or two Mobotix (M12, M15 or M24) webcams. Each webcam ran autonomously on a bat-tery powered by a solar panel. Images with a resolution of about 3 megapixels (2048 × 1536 pixels) were taken at hourly intervals during daylight, occasionally at 15‐min intervals during highly active periods (eg, during intense snowmelt). The images were sent via a
FIGURE 4 Time series of annual three‐
dimensional surface velocities of the terminal parts the three rock glaciers, measured by differential global navigation satellite system (dGNSS) since 2007 (average of n selected markers). The two sets of curves displayed for the Gugla rock glacier represent the velocities of the northern and the southern part of the front, respectively. The two crosses correspond to the fall of marked boulders on both parts of the front
FIGURE 5 Webcam image showing lightly disrupted winter snow
cover (arrows) that indicates the presence of the main shear horizon at the front of the Tsarmine rock glacier (webcam16, January 22 2017)
GSM (Global System for Mobile Communication) internet connec-tion to a server where the data were stored and accessible. The webcam system allowed remote live access but did not permit recording of movies. The first camera was installed in 2009 in Tsarmine and the last in 2015 there. Webcam characteristics are detailed in Table 1, and image examples for each webcam are shown in Figure 3.
Installation of cameras is often challenging in steep and remote environments. The location must allow the camera to have a good con-nection to a GSM signal for the data to be regularly sent to the server, and the solar panel must receive enough sunlight to charge the battery. The location must also be safe from rock falls and avalanches. The cam-era should be close enough to the rock glacier front to provide images with a resolution sufficient to detect erosion events of small scale (< ca 3 m3). At Gugla and Tsarmine, one webcam was installed near the rock glacier front (cam07 at Gugla, cam01 at Tsarmine; Table 1) to visu-alize the general movement there. These webcams provided a good view on the erosion processes at the front but did not cover the entire area of interest. Therefore, additional webcams were installed later to view the debris slope adjacent to the rock glacier fronts (cam14 was installed in June 2013 at Gugla and cam16 in May 2015 at Tsarmine). At Dirru, the lack of suitable positions led to only one webcam being installed quite far from the rock glacier front, which offered a good gen-eral view of the area of interest but not of small‐scale erosion processes. Geomorphological information was extracted from the image series by examining them visually and stored in a database. Between
two or more consecutive images, changes such as the loss or move-ment of rock debris at the surface of the front were identified as ero-sion events and attributed to an eroero-sion process, as illustrated in Figure 6. The high frequency of image acquisition also allowed moni-toring of erosion events through time. Although some images could not be analyzed due to poor weather conditions (especially fog) and some images were missing because of technical issues (eg, low bat-tery), almost the whole study period was covered for each site by sev-eral images taken every day by at least one webcam (Table 1). The information collected was stored in tables where each observed change at the surface of the front of the rock glacier was recorded with its date and associated with a type of erosion process. For each day of the study period, additional information such as general weather conditions (ie, rainfall, snowfall, sun, clouds) or the characteristics of the snow cover at the surface of the front (ie, full snow cover, partial snow cover, snow‐free) was also inferred from the images and recorded in the database.
4
|R E S U L T S
4.1
|Erosion processes
Four different types of events were detected from the images, based on categories of erosion processes (eg,34,35), namely rock fall, debris
slide (without rupture), widespread superficial flow and concentrated flow (including debris flows).
TABLE 1 Webcams and images characteristics
Site Name Date of installation Number of images (01.01.2013–31.12.2016) Distance to the front line (m) Focal length (mm) Covered area
Dirru Cam08 01.07.2011 17 636 407 65 Rock glacier back, front and debris slope
Gugla Cam07 28.06.2011 16 874 152 32 Rock glacier back and front
Cam14 13.07.2013 17 125 387 32 Rock glacier front and debris slope
Tsarmine Cam01 15.10.2009 16 819 78 43 Rock glacier back and front
Cam16 07.05.2015 9911 172 38 Rock glacier front and debris slope
FIGURE 6 Webcam images showing evidence of erosion (arrows) at the front of Gugla rock glacier between June 17 2013 (left) and June 18 2013
(right). Deep incision due to concentrated flow is indicated by the two large red arrows and by the red dotted line on the right‐hand image. The smaller red arrows point to more localized areas where smaller changes can be detected between the two images, mainly related to small rock falls (two arrows close to the front line) and to the accumulation of debris eroded from the front (lower small arrow) [Colour figure can be viewed at wileyonlinelibrary.com]
Rock fall consisted of the fall of one or a few boulders from the rock glacier front, and was the most common erosional process observed. On the image series, rock fall events were detected when boulders moved from their initial location (Figure 6), while no other notable changes affected the surrounding area. Some missing boulders could be identified downslope in the images. The fall of smaller debris (< ca 30 cm in diameter) was usually not recorded because the associ-ated morphological changes could not be detected. The fall or collapse of parts of the permafrost body (frozen debris) was not observed.
Debris slide consisted of the translational downward motion of a restricted mass of debris (ca 3–30 m3) in the unfrozen superficial layer
of the rock glacier front. Sliding events were characterized by the absence of rupture or fall and were usually traceable on several con-secutive images and often over several days. Debris slides commonly led to a single or multiple ruptures, which then sometimes triggered localized rock falls. Many slides were therefore associated with rock fall activity, but rock fall events were not necessarily preceded by a slide. Sliding of frozen debris (permafrost) was not observed.
Widespread superficial flow was associated with low‐discharge water circulation at the surface or within the top few decimeters of the rock glacier front and downstream debris slope. It was typically an areal process (unconcentrated wash), but did sometimes lead to the development of small linear erosional features, such as rill wash and small mudflows. Superficial flow events were identified by a wet ground surface (during snowmelt or rainfall), and were commonly asso-ciated with the small movement of debris over the entire surface of the rock glacier front. In addition, water temporarily flowing on the surface of the front and the downstream debris slope was commonly identified by the presence of small superficial mudflows (Figure 7: 1a, 2a and 3a). Concentrated flow was exclusively associated with significant linear erosion. It was characterized by both greater water input and incision than superficial flow, mobilizing up to several hundred cubic meters of material, mainly from the debris slope downstream of the rock glacier front. Concentrated flow events always related to the occurrence of water springs, which were commonly easy to detect (Figure 7: 1b, 2b and 3b). When discharge was high enough to mobilize large amounts of sediment, concentrated flow could lead to the occurrence of debris flows. Such events were triggered on the rock glacier fronts or on the adjacent debris slopes and some traveled for several kilometers.
4.2
|Temporal occurrence of the processes
The temporal resolution of the observations allowed the seasonal behaviors of the four main erosion processes at Dirru, Gugla and Tsarmine to be assessed (Figure 8). Temporal patterns of the different processes were similar for each site, though slight differences in timing occurred between some years and sites.
Generally, erosion events did not occur when the rock glacier front was covered by snow (Figure 8). Only a few rock fall events were observed when snow cover was present, mostly in winter 2015/16 and December 2016. Erosion events were more fre-quently observed once snow cover began to melt in March or April. Initially, rock falls and debris slides involving small amounts of sed-iments were more common. Widespread superficial flow began later, typically in late April or May, once the snow had melted from
at least part of the front. Initiation of these three processes was restricted to the rock glacier front (up to the front line). They were most frequent between May and June, when the front was snow‐ free but snowmelt was still active further upslope. Concentrated flow events were associated with intense snowmelt in the upslope catchment, sometimes coinciding with heavy or repeated rainfall. They were triggered either on the debris slope downstream of the rock glacier front or on the front itself, depending on the loca-tion of water springs.
The frequency of all processes tended to decrease gradually from the beginning of July until autumn. In July and August, rock fall was the most frequent process, and its occurrence was not necessarily associated with rainfall. Debris slides and widespread superficial flow events occurred more sporadically between July and October than during snowmelt and were usually associated with rainfall events. Concentrated flow events were rare during summer and autumn and required heavy or repeated rainfall. In autumn, the frequency of erosion events decreased substan-tially as no heavy rainfall occurred. Erosion event initiation almost completely ceased with the refreezing of the active layer and the estab-lishment of winter snow cover, generally in October or November.
5
|I N T E R P R E T A T I O N
5.1
|Causes of the erosion processes
Two main causes of erosion can be highlighted based on the timing and physical characteristics of each process: (i) rock glacier advance and (ii) water inputs (Figure 9). The rock glacier movement, characterized by a higher displacement rate at the surface than at depth (deformational flow), causes the frontal slope to steepen, which increases the shear stress on the sediment particles reposing on the front. When a certain threshold is passed, the rock particles which are not cemented by ice begin to reorganize by gravity.35This action causes individual or cas-cading gravitational movements in the form of isolated rock falls, which usually occur randomly in space and time. The increasing ice content resulting from the active‐layer freezing in late autumn and winter most likely cements the rock particles and therefore essentially prevents rock falls during the cold season, even though the rock glacier still moves.
The decrease in cohesive strength between rock particles, which relates to the second main cause of erosion, can be triggered by water input from (i) rainfall and/or snowmelt on the rock glacier front, (ii) thaw of the active layer or upper permafrost, or (iii) lateral groundwater flow in saturated or unsaturated horizons. In the two first cases, the ground humidification is commonly shallow and mainly results in the occurrence of rock falls, debris slides and widespread superficial flow events. Mud-flows may also be generated. Conversely to rock falls, debris slides and superficial flow events affect larger areas on the frontal slope and occur only when water is added. Subsurface saturation of the ground by shallow (suprapermafrost) or deeper (intra‐ or subpermafrost) groundwater flow can initiate water springs at the surface of the rock glacier front or on the debris slope downstream (Figure 7: 1b, 2b and 3b). The consecutive concentrated flow commonly triggers linear regressive erosion, beginning downstream from the outflow locations. Depending on the total dis-charge and the location of the activated water spring, the regressive
erosion can mobilize large volumes of sediment and ascend to the rock glacier front (see video at http://www.youtube.com/watch?v= 0k8OYEvHD_Y), where it may be constrained by the exposure of resis-tant frozen ground (permafrost). At the three sites, the initiation of con-centrated flow events required specific weather conditions that provided adequate water infiltration rates into the ground upslope of the rock glacier fronts. Such conditions occurred relatively infrequently and resulted from intense snowmelt phases at the catchment scale, heavy or recurring rainfall, or both.
5.2
|General erosional behavior during an annual
cycle
During an annual cycle, the timing of erosion events was similar at all three study sites. The erosion activity at the front of rock glaciers
can be separated into three main periods: winter, melting, and summer (Figures 8–10).
1. During winter (roughly from November to March), when the active layer was entirely frozen and the front was covered by snow, very few erosion events were observed. Some rock falls were detected, however, for instance during winter 2016, at all three sites. These may be explained by the lack of snow between December 2015 and February 2016, which exposed the surface to direct solar radiation in the afternoon. Superficial thaw could therefore have occurred on the fronts, facilitating small rock fall events. Rapid rock glacier movement may also enhance rock fall activity during winter. When rock glaciers move faster than 10 or 20 m/y, as recorded during the 2008–11 destabilization phase affecting the Grabengufer rock glacier in Mattertal, Switzerland,
FIGURE 7 Webcam images showing examples of erosional activity resulting from water flow at the front of Dirru (top), Gugla (middle) and
Tsarmine rock glaciers (bottom). Mudflows due to superficial flow (yellow arrows) are shown in 1a for Dirru (webcam08, may 11 2015), 2a for Gugla (webcam14, may 20 2014) and 3a for Tsarmine (webcam16, June 3 2015). Water springs generating concentrated flow erosion (red arrows) are shown in 1b for Dirru (webcam08, June 10 2014), 2b for Gugla (webcam14, June 23 2016) and 3b for Tsarmine (webcam16, June 24 2016)
rock fall activity may occur through the cold season (Figure 11).27
Nevertheless, sediment erosion activity in winter can typically be considered negligible. However, the rock glacier still advances in winter, even when no erosion occurs, moving significant volumes of material forward and increasing the slope angle between the front line and the debris slope below.
2. The transitional melting period (between March and June) began with the onset of snowmelt on the rock glacier. During this period, erosion activity rapidly increased in response to both the steeper frontal slope (favoring gravitational instabilities) and the input of water, first from snowmelt and active‐layer thaw, and later from permafrost thaw. In the early melting period (March–May; Figure 10), snowmelt and the active‐layer thaw mainly triggered rock falls, rock slides and widespread superficial flow. Later (May–June, Figure 10), the removal of debris on the rock glacier front may expose the permafrost table at the surface or leave it protected
only by a thin layer of debris. Continuous adjustment of the perma-frost table occurs on the rock glacier front, and permaperma-frost thaw contributes to the initiation of further frequent but low‐intensity erosion events, which lead to a progressive decrease of the slope angle between the front line and the debris slope. In addition, when snowmelt affects the whole catchment area (usually in June), sub-stantial water infiltration rate may increase the discharge at the water springs and trigger concentrated flow events, usually charac-terized by larger erosion volumes. The timing, duration and inten-sity of the melting period varied from year to year but typically lasted several weeks between late March and late June (sometimes even early July). For all sites in all years, the highest erosion activity was in the melting period, due to the increased and unbalanced slope angle and availability of water inputs.
3. During summer (roughly from July to October), the active‐layer usually continued to deepen, but at a lower rate than earlier in
FIGURE 8 Summary of the timing of erosion processes at the front of Dirru, Gugla and Tsarmine rock glaciers for the years 2014–16. Debris slides
and superficial flow events are associated in one line as they shared similar triggering conditions. Information about the snow cover on the front of the rock glacier (horizontal light grey lines) and the interpreted occurrence of rainfall events (vertical dark grey lines) are indicated. Different colors divide each year into three main periods differing in terms of erosion activity: Winter (blue), melting (yellow to red) and summer (green). Periods with no data are due to technical issues at Dirru and Gugla, and the absence of webcam from which the debris slope could be observed before 2015 in Tsarmine
the season. The gradual changes in the frontal slope geometry induced by the rock glacier advance are directly balanced by erosion events (mainly rock fall). Hence, the slope angle between the front line and the debris slope remains almost unchanged over summer. In general, the frequency of erosion events decreases gradually until the ground freezes again in October or November. Rainfall is the main source of ground moisture and can temporarily increase erosion intensity via widespread superficial flow, debris slides and rock falls. Heavy or recurring rainfall events may allow sufficient water infiltra-tion to episodically reactivate or increase the discharge of local water springs and trigger high‐magnitude regressive erosion events (concentrated flow).
6
|D I S C U S S I O N
6.1
|Differences between the sites and controlling
factors.
The most frequently observed erosion mechanism at all sites was rock fall, followed by debris slide and superficial flow, which require wet conditions near the ground surface. All three processes were observed each year at all study sites and usually involved the transfer of small amounts of sediments per event (typically ca 3–30 m3) over relatively
small distances (up to ca 50 m). They were therefore high‐frequency but low‐magnitude events (Figure 9). Conversely, erosion triggered by concentrated flow was less frequent but mobilized much larger vol-umes of sediment per event (ca 100 to more than 1000 m3). The
groundwater flow discharge must have been intense enough to induce a major increase in water content within the sediments on the debris slope downstream from the rock glacier front. Such conditions only developed when intense snowmelt affected a large part of the catch-ment area, during substantial or repeated rainfall over a period of up to a few days, or a combination of both. In the latter case, the amount of rainwater necessary to trigger a major debris flow was less impor-tant, because the debris had already been wetted from snowmelt.
Small differences in the timing or frequency of erosion events existed between the sites, mostly related to weather conditions con-trolling rainfall and snowmelt, which can be highly variable in space. For example, thunderstorms may affect very small areas, whereas snow cover development can differ at a local and regional scale depending on wind exposure and location within the mountain range. Other parameters such as the material erodibility on the frontal slope also play a role. Erosion from widespread superficial flow was more frequently observed at Gugla and Dirru than at Tsarmine because their fronts contain finer sediments, favoring the development of small mudflows. In addition, the occurrence of debris flows was always related to concentrated flow and depended on site‐specific factors.
Variations in local groundwater circulation and the number and location of water springs influenced the occurrence of concentrated flow. These local variations explain some of the differences observed between sites, such as the frequency and magnitude of debris flows. At Tsarmine, there was only one visible spring close to the front (Figure 7: 3b), which probably corresponded to intra‐ or suprapermafrost flow and was rarely active. It was usually characterized by a low discharge, inhibiting the development of high‐magnitude debris flows. At Dirru, one main spring was identified each year during intense snowmelt periods. It seemed to be linked to intrapermafrost
FIGURE 9 Summary of erosion processes in relation to their cause, type, timing and intensity during the three main periods (colors and circle). The
graphs on the right show the timing and intensity of the different processes and summarize information gathered from all sites for all documented years
flow and sometimes coincided with emergence of low‐discharge suprapermafrost water (Figure 12). When the discharge was highest at the main spring, the concentrated flow triggered debris flows, which sometimes reached the main channel of the Geisstriftbach torrent (Figure 7: 1b). At Gugla, several water springs were often visible either
during intense snowmelt periods and/or during persistent rainfall events. Most of the discharge from these springs was concentrated on the debris slope downstream from the rock glacier front and seemed to gather water from supra‐ and subpermafrost flows (Figure 7: 2b). No evidence of intrapermafrost flow was observed at Gugla. The nota-bly larger catchment area at Gugla than at the two other sites may have favored higher discharge at the springs. Concentrated flow leading to debris flows was thus most frequent at Gugla, where the largest magni-tude events were observed (several thousand cubic meters). At Dirru, only small to medium‐sized debris flows were recorded (up to several hundred cubic meters), and rarely reached the main valley. At Tsarmine, only very small events restricted to the upper part of the debris slope were observed (several tens of cubic meters).
6.2
|Significance of rock glaciers as sediment
sources
Despite differences related to local and external factors, the erosion processes observed at the three sites were the same and their occur-rence followed similar temporal patterns. At each site, the fre-quency–magnitude of the erosion processes was closely related both to the presence of permafrost and to the rock glacier dynamics. The presence of frozen ground explains the lack of erosion during winter when the active layer was frozen, and the high‐frequency events dur-ing snowmelt and active‐layer thaw due to the frequent supply of water to the sediments. Frozen ground also controls the magnitude of the erosion events as it limits the volume of thawed, erodible ment. These temporal considerations are also important in other sedi-ment accumulations such as moraines and talus slopes in mountain permafrost environments. However, the rock glacier dynamics gener-ate instabilities on the frontal slope, which allow a high frequency of rock fall events. The progressive steepening between the front line and the debris slope downstream promotes high erosional activity dur-ing the meltdur-ing season. Finally, the magnitude of rock glacier move-ment controls the continuous renewal of sedimove-ment available at the front. At the three study sites, the high surface velocity rates ensure a substantial availability of new sediment for further gravitational transfer by debris flows.30Torrents fed by rapidly moving rock glaciers
can therefore be considered as supply‐unlimited at an annual time scale, given that the sediment supply is renewed and sometimes even increased each year (eg,13).
6.3
|Limitations of the method
The results of this study relied on observations from webcam images, an approach that has several inherent limitations. First, the temporal coverage of the images was not always optimal. Erosion events at night may not leave visible traces and may therefore not all be recorded. Some images were missing or unusable because of technical problems or fog. The inventory of erosion processes presented here is therefore not exhaustive due to these data gaps. Moreover, some webcams were installed only recently to complement existing visual information. For example, at Tsarmine, webcam16 was set up in May 2015 (Table 1). Before this, images were only available from web-cam01, which only recorded some parts of the rock glacier front. At
FIGURE 10 Schematic vertical profiles of a rock glacier front
undergoing different periods of erosion activity within a 1‐year cycle (1‐3). In winter (1) the rock glacier advances but no erosion occurs and the frontal slope steepens. During the melting period (2a and 2b), the steep slope angle associated with water input from snowmelt, active‐ layer thaw (2a) and permafrost thaw during the readjustment of the permafrost table (2b) enhances erosion activity, which results in the adjustment of the frontal slope. During summer, erosion is less intense and balances approximately rock glacier advance (3)
Gugla, the installation of webcam14 in June 2013 improved observa-tions of some processes, particularly those related to superficial or concentrated flow (Table 1).
The detection of erosion events and their classification into pro-cess types was completed visually. Therefore, some small‐scale events may not have been recorded. The classification of such events is based on several visual clues. Hence, the inventory of the events and their categorization rely on the subjective judgment of the observer. How-ever, the possible misinterpretation or nondetection of some events and the lack of usable images for short periods was somewhat counterbalanced by the relatively long, 4‐year study period.
7
|C O N C L U S I O N
Quasi‐continuous sequences of webcam images revealed four main types of erosion processes at the front of three rapidly moving rock glaciers: rock fall, debris slide, superficial flow and concentrated flow. The main causes of erosion were either changes to the rock glacier front slope angle due to deformational flow, or the increase in moisture content of the rock debris constituting the fronts. Rock fall, debris slide and superficial flow were characterized by low magnitudes but rela-tively high frequencies, which varied between three main periods annually: winter (November to March), melting (March to June) and summer (July to October). Winter was characterized by a very low
FIGURE 11 Photographs showing the evolution of winter debris production from rock fall activity at the front of the Grabengufer rock glacier
between 2010 and 2013 in relation to surface velocity. Above a velocity of about ~10 m/y (a, b) winter ground freezing was insufficient to prevent erosion at the front. Below this velocity (c, d), rock fall during winter almost ceased [Colour figure can be viewed at wileyonlinelibrary.com]
FIGURE 12 Photograph of the front of the Dirru rock glacier during a
melting period (June 17 2013). Substantial removal of debris during the first part of the melting period exposed the permafrost table at the ground surface. Intense snowmelt activated two water springs visible on the image [Colour figure can be viewed at wileyonlinelibrary.com]
frequency of erosion events because the ground is usually entirely fro-zen. Rock glaciers continue to creep forward in winter, steepening their frontal slopes. Most erosion events were observed during the melting period in association with the increased front slope angle and water input from snowmelt, active‐layer thaw and permafrost thaw. In summer, rock glacier movement induces gradual changes of the front slope angle, which are usually balanced by rock falls.
Occasional inputs of water from rainfall can temporarily increase the frequency of erosion events. Groundwater emerging in the debris slope or at the front of the rock glacier may cause significant regressive erosion (concentrated flow) and sometimes trigger debris flows. Con-centrated flow is generated by water inputs from intense snowmelt and/or heavy or repeated rainfall, and is generally characterized by low frequency (a few days per year) and relatively high magnitude, depending on local factors such as the potential for groundwater discharge.
Compared to other high‐mountain sediment sources, the fronts of rapidly moving rock glaciers connected to torrential channels show high erosion and sediment transfer activity. Rock glacier movement favors the initiation of erosion by continuously changing the frontal slope, which causes instabilities. It also ensures a continuous renewal of sediments available for erosion. Erosion intensity can vary substan-tially within and between years, controlled by weather conditions and by the kinematical behavior of the rock glacier. Although quantifica-tions of erosion rates are still lacking, the observaquantifica-tions showed that higher rock glacier velocities can be expected to increase the sediment renewal rate at their fronts and favor sediment transfer between rock glaciers and torrential systems. The recent increase in rock glacier velocities therefore implies substantial modification of the sediment cascade by locally increasing sediment availability in the headwaters of high‐mountain torrential catchments.
A C K N O W L E D G E M E N T S
The on‐site installation and maintenance of the webcams were carried out with the participation and the help of numerous individual, who we thank. We also acknowledge the municipalities of Evolène and St. Niklaus as well as the Service des forêts et du paysage du Canton du Valais for the logistical and financial support which strongly facilitated the fieldwork operations. The constructive comments and remarks from the editor, the associate editor and the two referees are gratefully acknowledged, along with the help received from Prof J. Murton and Dr B. O'Neill with proofreading of the manuscript.
O R C I D
Mario Kummert http://orcid.org/0000-0002-1815-883X
R E F E R E N C E S
1. Caine N. The geomorphic processes of the alpine environment. In: Ives JD, Barry RG, eds. Arctic and Alpine Environments. London: Methuen; 1974:721‐748.
2. Bracken LJ, Turnbull L, Wainwright J, Bogaart P. Sediment connectivity: a framework for understanding sediment transfer at multiple scales. Earth Surf Proc Land. 2015;40(2):177‐188.
3. Kaufmann V, Ladstädter R, Kienast G. 10 years of monitoring of the Doesen rock glacier (Ankogel Group, Austria)—a review of the research activities for the time period 1995‐2005. In: Petrovic D, ed. Proceedings
of the 5th Mountain Cartography Workshop, 29 March–April 2006. Slovenia: Bohinj; 2007:129‐144.
4. Roer I, Haeberli W, Avian M, et al. Observations and considerations on destabilizing active rock glaciers in the European Alps. In: Kane DL, Hinkel KM, eds. Proceedings of the 9th International Conference on Per-mafrost, June 29–July 3, 2008. Fairbanks, Alaska: Institute of Northern Engineering, University of Alaska Fairbanks; 2008:1505‐1510. 5. Delaloye R, Lambiel C, Gärtner‐Roer I. Overview of rock glacier
kine-matics research in the Swiss Alps. Geogr Helv. 2010;65(2):135‐145. 6. PERMOS. Permafrost in Switzerland 2010/2011 to 2013/2014. In:
Noetzli J, Luethi R, Staub B, eds. Glaciological Report (Permafrost) No. 12–15. of the Cryospheric Commission of the Swiss Academy of Sciences; 2016.
7. Daanen RP, Grosse G, Darrow MM, Hamilton TD, Jones BM. Rapid movement of frozen debris‐lobes: implications for permafrost degrada-tion and slope instability in the south‐central Brooks Range, Alaska. Nat Hazards Earth Syst Sci. 2012;12(5):1521‐1537.
8. Sorg A, Kääb A, Roesch A, Bigler C, Stoffel M. Contrasting responses of central Asian rock glaciers to global warming. Sci Rep. 2015;5(8228):8228
9. Kääb A, Strozzi T, Sorg A, Stoffel M. Variations in rockglacier speed in the Tien Shan and their significance. In: Günther F, Morgenstern A, eds. XI International Conference on Permafrost – Book of Abstracts, 20–24 June 2016. Potsdam: Germany Bibliothek Wissenschaftspark Albert Einsteinh; 2016.
10. D'Agostino V, Bertoldi G. On the assessment of the management prior-ity of sediment source areas in a debris‐flow catchment. Earth Surf Process Landf. 2014;39(5):656‐668.
11. Hürlimann M, Copons R, Altimir J. Detailed debris flow hazard assess-ment in Andorra: a multidisciplinary approach. Geomorphology. 2006;78(3–4):359‐372.
12. Stoffel M. Magnitude–frequency relationships of debris flows — a case study based on field surveys and tree‐ring records. Geomorphology. 2010;116(1–2):67‐76.
13. Bovis MJ, Jakob M. The role of debris supply conditions in predicting debris flow activity. Earth Surf Proc Land. 1999;24(11):1039‐1054. 14. Jakob M, Bovis M, Oden M. The significance of channel recharge rates
for estimating debris‐flow magnitude and frequency. Earth Surf Process Landf. 2005;30(6):755‐766.
15. Theule JI, Liébault F, Laigle D, Jaboyedoff M. Sediment budget moni-toring of debris‐flow and bedload transport in the Manival torrent, SE France. Nat Haz Earth Syst Sci. 2012;12:731‐749.
16. Croke J, Mockler S, Fogarty P, Takken I. Sediment concentration changes in runoff pathways from a forest road network and the resul-tant spatial pattern of catchment connectivity. Geomorphology. 2005;68(3–4):257‐268.
17. Fryirs K. (Dis)connectivity in catchment sediment cascades: a fresh look at the sediment delivery problem. Earth Surf Process Landf. 2013;38(1):30‐46.
18. Oggier N, Graf C, Delaloye R, Burkard A 2016. Integral protection con-cept “Bielzug”—Integrales Schutzkonzept Bielzug. In Conference Proceedings, INTERPRAEVENT 2016:525–534.
19. Wahrhaftig C, Cox A. Rock glaciers in the Alaska range. Geol Soc Am Bull. 1959;70(4):383‐436.
20. Haeberli W. Creep of mountain permafrost: internal structure and flow of alpine rock glaciers. Mitteilungen der VAW‐ETH Zürich. 1985;77:1‐172. 21. Kääb A, Reichmuth T. Advance mechanisms of rock glaciers. Permafrost
Periglac Process. 2005;16(2):187‐193.
22. Haeberli W, Hoelzle M, Kääb A, Keller F, Vonder Mühll D, Wagner S. Ten years after drilling through the permafrost of the active rock glacier Murtèl, eastern Swiss Alps: answered questions and new perspectives. In: Lewcowicz AG, Allard M, eds. Proceedings of the 7thInternational
Conference on Permafrost, June 23–27, 1998. Collection Nordicana. Vol.57 Yellowknife, Canada; 1998:403‐410.
23. Barsch D, Caine N. The nature of mountain geomorphology. Mt Res Dev. 1984;4(4):287‐298.
24. Gärtner‐Roer I. Sediment transfer rates of two active rockglaciers in the Swiss Alps. Geomorphology. 2012;167‐168.
25. Müller J, Gärtner‐Roer I, Kenner R, Thee P, Morche D. Sediment stor-age and transfer on a periglacial mountain slope (Corvatsch, Switzerland). Geomorphology. 2014;218:35‐44.
26. Lugon R, Stoffel M. Rock‐glacier dynamics and magnitude–frequency relations of debris flows in a high‐elevation watershed: Ritigraben, Swiss Alps. Glo Planet Change. 2010;73(3–4):202‐210.
27. Delaloye R, Morard S, Barboux C, et al. Rapidly moving rock glaciers in Mattertal. In: Graf C, ed. Mattertal– ein Tal in Bewegung. Publikation zur Jahrestagung der Schweizerischen Geomorphologischen Gesellschaft, 29 June – 1 July 2011. St. Niklaus, Switzerland. Birmensdorf, Eidg: Forschungsanstalt WSL; 2013:113‐124.
28. Bauer A, Paar G, Kaufmann V. Terrestrial laser scanning for rock gla-cier monitoring. In: Phillips M, Springman SM, Arenson LU, eds. Permafrost. Proceedings of the Eighth International Conference on Permafrost, 21–25 July 2003. Vol.2003 Zurich, Switzerland. Lisse, The Netherlands: A.A. Balkema:55‐60.
29. Avian M, Kellerer‐Pirklbauer A, Bauer A. LiDAR for monitoring mass movements in permafrost environments at the cirque Hinteres Langtal, Austria, between 2000 and 2008. Nat Haz Earth Syst Sci. 2009;9:1078‐1094.
30. Kummert M, Delaloye R. Quantifying sediment transfer between the front of an active alpine rock glacier and a torrential gully. In: Jasiewicz
J, Zwolinski Z, Mitasova H, Hengl T, eds. Geomorphometry for Geosciencesnań. Poznan: Adam Mickiewicz University in Poznan ‐ Institute of Geoecology and Geoinformation, International Society for Geomorphometry; 2015:193‐196.
31. Kneisel C, Rothenbühler C, Keller F, Haeberli W. Hazard assessment of potential periglacial debris flows based on GIS‐based spatial modelling and geophysical field surveys: a case study in the Swiss alps. Permafr Periglac Process. 2007;18(3):259‐268.
32. Arenson L, Hoelzle M, Springman S. Borehole deformation measure-ments and internal structure of some rock glaciers in Switzerland. Permafr Periglac Process. 2002;13(2):117‐135.
33. Bearth P. Geologischer Atlas der Schweiz 1: 25 000, Karte 43. Blatt 1328. Swisstopo: Randa.
34. Varnes DJ. Slope movement types and processes. In: Schuster R, Krizak R, eds. Landslides. Washington, DC: National Academy of Sciences Transportation Research Board; 1978:11‐33.
35. Easterbrook DJ. Surface Processes and Landforms. 2nd ed. Prentice Hall: Upper Saddle River, NJ; 1999.
How to cite this article: Kummert M, Delaloye R, Braillard L. Erosion and sediment transfer processes at the front of rapidly moving rock glaciers: Systematic observations with automatic cameras in the western Swiss Alps. Permafrost and Periglac Pro-cess. 2018;29:21–33.https://doi.org/10.1002/ppp.1960